GAD and related anxiety disorders in youth are chronic and highly impairing. Cognitive-behavioral therapy (CBT), and other active forms of psychotherapy (e.g. education, support), are associated with reductions in anxiety (Barrett et al 1996;Kendall et al 1997;Last et al 1998;Silverman et al 1999);however, approximately 44% of anxious children treated with CBT do not improve (James et al 2005). Understanding predictors and correlates of treatment response in child anxiety will allow us to: (1) target treatments to children most likely to benefit, (2) refine treatments by focusing on components shown to be associated with treatment response, and (3) develop new treatments tailored to CBT non-responders. We propose that child anxiety disorders, as exemplified by GAD (Rapee, 2002), are associated with a "vigilance-avoidance" pattern characterized by problems in monitoring and evaluating emotional information and modifying emotional reactions. This pattern can be reinforced by controlling and critical parenting behaviors, parental psychopathology, and negative interactions with peers. CBT treatments for child anxiety target disruptions in emotion regulation by teaching children skills for identifying and managing negative emotion and providing opportunities to practice these skills during exposures. These skills are then presumed to generalize to social interactions in daily life settings outside the clinic, including interactions with parents and peers. Although improvements in affective functioning in the social context are believed to play a role in anxious children's positive response to CBT, research has not yet clearly demonstrated links between children's or parents' affective behaviors and their response to treatment. We will examine how children's emotion regulation in the social context and their relationships with parents and peers (1) predict initial and long-term response to CBT treatment (individually and compared to an active comparison Child Centered treatment) and (2) change across the course of treatment. We will rely on two ecologically valid methods for assessing affective behaviors in naturalistic contexts: (a) Behavioral Observation and (b) Ecological Momentary Assessment.